I’ve had some luck with giving the LLM an overview of what I want the final version to do, but then asking it to perform smaller chunks. This is how I’d approac

I’ve had some luck with giving the LLM an overview of what I want the final version to do, but then asking it to perform smaller chunks. This is how I’d approach it myself — I know where I’m trying to go, and will implement smaller chunks at a time. I’ll also sometimes ask it to skip certain functionality - leaving a placeholder and saying we’ll get back to it later.